Proton Drip-Line Calculations and the Rp-process

20 Pages·0.4 MB·English

by  B. A. Brown

Proton Drip-Line Calculations and the Rp-process
B. A. Brown, R.R.C. Clement, H. Schatz and A. Volya
Department of Physics and Astronomy
and National Superconducting Cyclotron Laboratory,Michigan State University,
East Lansing, Michigan 48824-1321,USA
W. A. Richter
Department of Physics, University of Stellenbosch,
Stellenbosch 7600,South Africa
2
0 One-protonandtwo-protonseparation energiesarecalculatedforproton-richnucleiintheregion
0 A=41−75. The method is based on SkyrmeHartree-Fock calculations of Coulomb displacement
2 energies of mirror nuclei in combination with the experimental masses of the neutron-rich nuclei.
n The implications for the proton drip line and the astrophysical rp-process are discussed. This is
a done within the framework of a detailed analysis of the sensitivity of rp process calculations in
J type I X-ray burst models on nuclear masses. We find that the remaining mass uncertainties, in

0 particularforsomenucleiwithN =Z,stillleadtolargeuncertaintiesincalculationsofX-rayburst
2 light curves. Further experimental or theoretical improvements of nuclear mass data are necessary
before observed X-ray burst light curves can be used to obtain quantitative constraints on ignition

1 conditions and neutron star properties. We identify a list of nuclei for which improved mass data
v would be most important.
3
5
0
1
0
2
0
/
h
t
-
l
c
u
n
:
v
i
X
r
a
1
I. INTRODUCTION
The masses for the proton-rich nuclei above A=60 have not yet been measured. However,they are important for
the astrophysical rapid-proton capture (rp) process [1] which follows a path in nuclei near N =Z for A =60−100.
The rp process is the dominant source of energy in type I X-ray bursts, and it determines the crust composition of
accreting neutron stars [2–6]. It may also be responsible for the p process nucleosynthesis of a few proton-richstable
nuclei in the A=74–98 mass range. In the absence of experimental masses for the proton-richnuclei, one often uses
the masses based upon the Audi-Wapstra extrapolation (AWE) method [7]. In this paper we use the displacement-
energy method [8–11] to obtain the proton-rich masses with the Skyrme Hartree-Fock model for the displacement
energies. The displacement energy is the difference in the binding energies of mirror nuclei for a given mass A and
isospin T:
D(A,T)=BE(A,T<)−BE(A,T>), (1)
z z
where T =| T< |=| T> |, BE(A,T<) is the binding energy of the proton-rich nucleus and BE(A,T>) is the
z z z z
binding energy of the neutron-rich nucleus. The displacement energy can be much more accurately calculated than
the individual BE in a variety of models since it depends mainly on the Coulomb interaction. In particular, we will
use the sphericalHartree-Fock model based upon the recent SkX set of Skyrme parameters [12], with the addition of
charge-symmetrybreaking(CSB),SkX [13]. Withthe additionofCSBthesecalculationsareableto reproducethe
csb
measured displacement energies for all but the lightest nuclei to within an rms deviation of about 100 keV [13]. In
the A=41−75 mass region the mass (binding energy) of most of the neutron-richnuclei are experimentally usually
known to within 100 keV or better (the only exception being 71Br for which we use the AWE). Thus we combine
the experimentalbinding energyfor the neutron-richnucleus BE(A,T>) together with the Hartree-Fockvalue for
z exp
D(A,T) to provide an extrapolation for the proton-rich binding energy:
HF
BE(A,T<)=D(A,T) +BE(A,T>) . (2)
z HF z exp
The method is similar to the one used by Ormand [10] for the proton-rich nuclei with A = 46−70. In [10] the
displacementenergiesarebaseduponshell-modelconfigurationmixingwhichincludesCoulombandCSBinteractions
withparametersforthesingle-particleenergiesandstrengthswhicharefittedtothismassregion. Inthepresentwork,
which covers the region A = 41−75, the displacement energies are based upon Skyrme Hartree-Fock calculations
with a global set of parameters which are determined from the properties of closed-shell nuclei and nuclear matter.
The CSB partofthe interactionhasone parameterwhichwasadjustedto reproducethe displacementenergiesinthe
A=48 mass region [13].
ThedisplacementenergiesforallbutthelightestnucleicanbereproducedwiththeconstantCSBinteractiongiven
in [13], and we use the same CSB interaction for the extrapolations to higher mass discussed here.
The calculations presented here are relevant for the masses of proton-rich nuclei via their connection with their
mirror neutron-rich analogues. We are not able to improve upon the masses of nuclei with N = Z, and as will be
discussed, the relative largeerrorswhich remain for the 64Ge and 68Se masses provide now the dominant uncertainty
in the rp-process calculations.
DetailsoftheHartree-Fockcalculationswillbediscussed,andacomparisonbetweenthecalculatedandexperimental
displacement energies for the A = 41−75 mass region will be made. Then the extrapolations for the proton-rich
masses and the associated one- and two-proton separation energies will be presented. The proton drip line which is
established by this extrapolation will be compared to experiment, and the nuclei which will be candidates for one-
andtwo-protondecaywillbediscussed. Finallyweexplorethesignificanceofthenewextrapolationfortherpprocess
in type I X-ray bursts.
II. DISPLACEMENT-ENERGY CALCULATIONS
The SkX interaction is used to carry out Hartree-Fock calculations for all nuclei in the range Z = 20− 38
csb
and N = 20−38. The binding energies are then combined in pairs to obtain theoretical displacement energies for
A=41−75 and T =1/2 to T =4:
D(A,T) =BE(A,T<) −BE(A,T>) . (3)
HF z HF z HF
The calculationis similar to those presentedin [13], but severalrefinements are made. The single-particle states in
proton-rich nuclei become unbound beyond the proton-drip line. In the nuclei we consider they are unbound by up
2
to about 2 MeV. Since 2 MeV is small compared to the height of the Coulombbarrier (about 6 MeV at a radius of 7
fm), the states are “quasi-bound” and have a small proton-decay width (on the order of keV or smaller). To obtain
the quasi-bound wave functions we put the HF potential in a box with a radius of 20 fm and a depth of 20 MeV. In
all cases we consider, the dependence of the results on the form of the external potential is negligible as long as the
radius is greater than about 10 fm and the potential depth is greater than about 10 MeV.
In [13] the occupation numbers of the spherical valence states were filled sequentially, and in this mass regionthey
always occur in the order f , p , f , p and g . We have improved on this scheme by carrying out an exact
7/2 3/2 5/2 1/2 9/2
pairing(EP)calculation[14]ateachstageoftheHFiteration. Theexactpairingmodelhasrecentlybeendiscussedin
[14]. TheEPmethodusesthesingle-particleenergiesfromtheHFcalculationtogetherwithafixedsetofJ =0,T =1
two-body matrix elements and gives the orbit occupations and the pairing correlation energy. The orbit occupations
arethen usedtogetherwith the HFradialwavefunctions to calculatethe nucleondensities whichgointo the Skyrme
energydensityfunctional. Thisprocedureisiterateduntilconvergence(about60iterations). Thepairingiscalculated
for protons and neutrons with the same set of two-body matrix elements taken from the FPD6 interaction for the
pf shell [15] and the Bonn-C renormalized G matrix for the matrix elements involving the g orbit [16]. For those
9/2
nuclei we consider here, the occupation of the g orbit is always small. It is known that deformed components of
9/2
the 2s-1d-0g shell are essential for the nuclear ground states above A = 76 as indicated by the sudden drop in the
energy of the 2+ state from 709 keV in 72Kr to 261 keV in 76Sr [17]. Thus we do not go higher than A = 76. In
addition, one cannot always use Eq. (2) above A = 76 since many of the masses of the neutron-rich nuclei are not
known experimentally.
Theresultsweobtainarenotverysensitivetothestrengthofthepairinginteractionandtheassociateddistribution
ofthenucleonsbetweenthepandf orbits,sincetheseorbitshavesimilarrmsradiiandsingle-particleCoulombshifts.
For example, a 20 percent change in the strength of the pairing interactionresults in displacement energy changes of
less than 20 keV. If pairing is removed, the displacement energies can change by up to about 100 keV. Thus, at the
level of 100 keV accuracy pairing should be included, but it is not a crucial part of the model.
A final refinement has been to add a Coulomb pairing contribution to the proton-proton J = 0 matrix elements.
The two-body Coulomb matrix elements were calculated in a harmonic-oscillatorbasis. The Coulomb pairing is then
definedasthedifferenceofthediagonalJ =0matrixelementsfromthe(2J+1)weightedaverage(whichcorresponds
to the spherical part of the Coulomb potential which is in the HF part of the calculation). The Coulomb pairing
matrix elements are 50-100 keV.
In Fig. 1 the calculated displacement energies (crosses) are shown in comparisonwith experiment (filled circles) in
cases where both proton- and neutron-rich masses have been measured and with the AWE (squares) in cases where
the mass of the proton-rich nucleus is based upon the AWE. The corresponding differences between experiment and
theory are shownin Fig.2 including the experimentalor AWE errorbars. It can be seenthat when the displacement
energy is measuredthe agreementwith the calculation is excellent to within an rms deviation of about 100 keV. The
most exceptional deviation is that for A=54 involving the 54Ni-54Fe mirror pair; a confirmationof the experimental
mass for 54Ni (which has a 50 keV error) would be worthwhile. The comparison based upon the AWE (squares)
shows a much larger deviation with typically up to 500 keV differences, but the AWE error assumed is sometimes
(but not always) large enough to account for the spread. The implication of this comparison is that the error in the
HF extrapolation of the displacement energies is probably much less than the error in the AWE of the displacement
energies. In particular, one notices in Fig. 1 in the region A=60−75 that the displacement energy based upon the
AWEshowsasmalloscillationwhichisnotpresentintheHFcalculationandwhichisnotpresentintheexperimental
data for A<60.
III. PROTON-RICH MASSES AND SEPARATION ENERGIES
The next step is to use Eq. (2) to calculate the binding energy of the proton-rich nuclei based upon the HF
calculation of the displacement energy together with the experimental binding energy of the neutron-rich nucleus
[7,18]. The only neutron-rich nucleus whose mass is not yet experimentally measured is 71Br for which we use the
AWE value. The binding energies for the HF extrapolations for the proton-richnuclei are given an error based upon
the experimental error of the neutron-rich binding energy folded in quadrature with an assumed theoretical error of
100 keV.
The HF extrapolatedset of binding energies for proton-richnuclei together with the experimentalbinding energies
for nuclei with N =Z and neutron-rich nuclei provides a complete set of values from which the one- and two-proton
separation energies are obtained. The masses for the N = Z nuclei 66As, 68Se, 70Br, are not measured and we use
the AWE value. The mass for 74Rb has a relatively large experimental error.
3
Results for the one- and two-proton separation energies are shown in Fig. 3. The first line in each box is the one-
proton separation energy (and the associated error)based upon the AWE with the associated error. The second line
is the one-protonseparationenergy baseduponthe HF extrapolation,and the third line is the two-protonseparation
energy based upon the HF extrapolation. The error in the separationenergies is the errorfor the binding energies of
the parent and daughter nuclei folded in quadrature.
The double line in Fig. 3 is the proton-drip line beyond which the one-proton separation energy and/or the two-
proton separation energy becomes negative. However, due to the Coulomb barrier, some of the nuclei beyond the
proton-driplinemayhavelifetimeswhicharelongenoughtobeabletoobservetheminradioactivebeamexperiments.
The observation of 65As in the experiment of Blank et al. [19] excludes half-lives which are much shorter than 1 µs
which indicates that it is unbound by less than 400 keV. The identification of 65As as a β-emitter by Winger et al.
[20]togetherwith the non-observationofemitted protonsby Robertsonetal.[21]indicates thatit is unboundby less
than 250 keV. Both limits are compatible (within error) with the HF results given in Fig. 3. The non-observation of
69Br in the radioactive beam experiments of Blank et al. [19] and Pfaff et al. [22] means that its lifetime is less than
24 nsec which implies that it is proton unbound by more than 500 keV [22]. This is compatible with the HF result
shown in Fig. 3. The non-observation of 73Rb in the experiments of Mohar et al. [23], Jokinen et al. [24], and Janas
et al. [25] gives an upper limit of 30 nsec for the half-life which implies that 73Rb is proton unbound by more than
570 keV, again in agreement (within error) of the present HF result. Thus all of the current experimental data are
consistent with our calculations.
The proton-drip line has not yet been reached for most Z values. Beyond the proton-drip line there are several
candidatesfornucleiwhichshouldbeexploredforone-protonemission: 54Cu,58Ge,64As,68Br,69Br,72Rband73Rb.
The most promising candidates for the illusive diproton emission (in addition to 48Ni [8,10]) are 64Zn, 59Ge, 63Se,
67Kr and 71Sr. Estimated lifetime ranges for these diproton decays are given by Ormand [10].
IV. IMPLICATIONS FOR THE RP PROCESS
The rp process beyond Ni plays a critical role during hydrogen burning at high temperatures and densities on the
surface of accreting neutron stars in X-ray bursters and X-ray pulsars [2–6]. Nuclear masses are among the most
important input parameters in rp-process calculations, as they sensitively determine the balance between proton
capture and the inverse process, (γ,p) photodisintegration. It is this (γ,p) photodisintegration that prevents the rp
process from continuing via proton captures, once a nucleus close to the proton drip line is reached. This nucleus
becomes thena ”waitingpoint” asthe rpprocesshas to proceedatleastin part,viathe slowβ+ decay. The effective
lifetimeofthewaitingpointsintherpprocessdeterminestheoverallprocessingtimescale,energygeneration,andthe
finalabundance distribution. At a waiting point nucleus (Z,N), a local(p,γ)-(γ,p) equilibrium is establishedwith the
following isotones (Z+1,N), (Z+2,N). The effective proton capture flow destroying waiting point nuclei and reducing
their lifetime is then governed by the Saha equation and the rate of the reaction leading out of the equilibrium.
Becauseoftheodd-evenstructureofthe protondripline2caseshavetobedistinguished[3]. Fortemperaturesbelow
≈ 1.4 GK equilibrium is only established with the following isotone (Z+1,N). In this case, the destruction rate of the
waiting pointnucleus via protoncapturesλ is determined by the Sahaequationandthe protoncapture rate
(Z,N)(p,γ)
on the following isotone (Z+1,N). The total destruction rate of the waiting point nucleus (Z,N) is then given by the
sum of proton capture and β-decay rates:
2π¯h2 3/2 G (T) Q
λ=λ +Y2ρ2N2 (Z+1,N) exp (Z,N)(p,γ) <σv > (4)
β p A(cid:18)µ kT(cid:19) (2J +1)G (T) (cid:18) kT (cid:19) (Z+1,N)(p,γ)
(Z,N) p (Z,N)
λ is the β-decay rate of nucleus (Z,N), Y the hydrogen abundance, ρ the mass density, J the proton spin, G
β p p (Z,N)
the partition function of nucleus (Z,N), T the temperature, µ the reduced mass of nucleus (Z,N) plus proton,
(Z,N)
Q the proton capture Q-value of the waiting point nucleus, and <σv > the proton capture rate
(Z,N)(p,γ) (Z+1,N)(p,γ)
on the nucleus (Z+1,N). For higher temperatures local equilibrium is maintained between the waiting point nucleus
(Z,N)andthenexttwofollowingisotones(Z+1,N)and(Z+2,N).Inthiscase,λ isgivenbytheSahaequation
(Z,N)(p,γ)
and the β-decay rate of the final nucleus λ , and the total destruction rate λ of the waiting point nucleus
(Z+2,N)β
becomes:
λ=λ +Y2ρ2N2 2π¯h2 3µ−3/2 µ−3/2 G(Z+2,N)(T) exp Q(Z,N)(2p,γ) λ (5)
β p A(cid:18) kT (cid:19) (Z,N) (Z+1,N)(2J +1)2G (T) (cid:18) kT (cid:19) (Z+2,N)β
p (Z,N)
In both cases, the destruction rate of a waiting point nucleus depends exponentially either on its one-proton capture
Q-value Q or two-proton capture Q-value Q . Nuclear masses therefore play a critical role in
(Z,N)(p,γ) (Z,N)(2p,γ)
determining the rp-process waiting points and their effective lifetimes.
4
It has been shown before that the most critical waiting point nuclei for the rp process beyond Ni are 64Ge, 68Se
and 72Kr [3]. With the exceptionof 56Ni and 60Zn, these nuclei are by far the longest-livedisotopes in the rp-process
path. The reasonfor those three nucleibeing the mostcriticalones is that withincreasingchargenumber the N =Z
line moves closer to the proton drip line and away from stability. Therefore, proton capture Q-values on even-even
N =Z nuclei, which are favoredin the rp process because of the odd-evenstructure of the protondrip line, decrease
with increasingchargenumber, while theβ-decay Q-values become larger. 64Ge, 68Se and72Krhappen to be located
in the ”middle”, where proton capture Q-values are already low enough to suppress proton captures and allow β
decay to compete, but at the same time β-decay Q-values are still small enough for half-lives to be long comparedto
rp-process time scales. The critical question is to what degree proton captures can reduce the long β-decay lifetimes
of 64Ge (63.7 s half-life), 68Se (35.5 s half-life) and 72Kr (17.2 s half-life). As Eqs. 4 and 5 show, the answer depends
mainly on the one- and two-protoncapture Q-values. Unfortunately, experimental data exist for none of the relevant
Q-values. The only available experimental information are upper limits of the one-proton capture Q-values of 68Se
and 72Kr from the non-observation of 69Br [19,22] and 73Rb [23–25], and the lower limits on the one-proton capture
Q-value on 65As from its identification as a β-emitter in radioactive beam experiments (see Sec. III). While these
data provide some constraints,accurate Q-values are needed for the calculations and have to be predicted by theory.
The new masses calculated in this work cover exactly this critical mass range, and provide improved predictions for
alltherelevantQ-valuesintheA=64–72massregion(seeFig.3). AsdiscussedinSec.III,allofournewpredictions
are compatible with the existing experimental limits.
To explore the impact of the new mass predictions on rp-process models, we performed calculations with a 1-D,
one zone X-ray burst model [26,6]. Ignition conditions are based on a mass accretionrate of 0.1 times the Eddington
accretionrate,aninternalheatfluxfromtheneutronstarsurfaceof0.15MeV/nucleon,anaccretedmattermetallicity
of 10−3 and a neutron star with 1.4 solar masses and 10 km radius.
In principle, proton separation energies can influence the reaction flow in two ways. First, they affect the forward
to reverse rate ratios for proton capture reactions and the local (p,γ)-(γ,p) equilibria through the exp(Q/kT) term
in the Saha equation (in Eqs. 4 and 5). This leads to an exponential mass dependence of the waiting point lifetimes.
Second,theoreticalpredictions ofreactionrates<σv >(in Eq.4)depend alsoonthe adoptedQ-values. Inthis work
wechoosetotakeintoaccountbotheffects. ToexploretheimpactofQ-valueuncertaintiesonprotoncapturereaction
rate calculations we use the statistical model code SMOKER [3]. Even though the nuclei in question are close to
the proton drip line a statistical approachis justified in most cases because reaction rates tend to become important
only for larger Q-values when a local (p,γ)-(γ,p) equilibrium cannot be established. Then the level density tends to
be sufficient for the statistical model approach. Based on the new reaction rates we then use our new Q-values to
recalculate (γ,p) photodisintegration rates via detailed balance as discussed in [3].
For the relevant temperature range between 1-2 GK, our new proton capture reactionrates vary in most cases not
morethanafactoroftwowithinthe exploredmassuncertainties. Anexceptionamongthe relevantreactionratesare
theprotoncapturerateson65,66As,69,70Br,and73,74Rb. Theseratesshowasomewhatstrongervariationoftypically
a factor of 4 – 6 as the associated proton capture Q-values are particularly uncertain. Fig. 4 shows two examples for
the Q-value dependence of statistical model reaction rates. Generally, a larger Q-value leads to larger rates, as the
higher excitation energy of the compound nucleus opens up more possibilities for its decay. For reference, Fig. 4 also
shows the rates listed in [3], which had been calculated using Q-values from the Finite Range Droplet Mass model
(FRDM1992) [27].
To disentangle the different effects of mass uncertainties quantitatively we performed test calculations in which
changes in masses were only taken into account in the calculation of the (γ,p) photodisintegration rates, while the
proton capture rates were kept the same. These test calculations lead to very similar luminosity and burst time
scale variations as presented in this paper. Discrepancies were at most 8% in the luminosity and 0.1% in the burst
timescale. This can be understood from Eq. 4 and 5. For example, a change of 1.37 MeV in the proton capture
Q-value changesthe 65As reactionrate and thereforethe lifetime of the 64Ge waiting point nucleus by a factor of 3–4
(see Fig. 4 and Eq. 4). However, the same 1.37 MeV Q-value change in the exp(Q/kT) term in Eq. 5 would result
in a lifetime change of 6 orders of magnitude (for a typical kT = 100 keV). We therefore conclude that the impact
of mass uncertainties on rp process calculations through changes in theoretical reaction rate calculations within the
statistical model is much smaller than the impact through changes in (p,γ)/(γ,p) reaction rate ratios.
The following calculations were performed with different assumptions on masses beyond the N = Z line from
Z = 30-38: SkX based on the mass predictions of this work, SkX-MIN with all proton capture Q-values set to the
lowest value, and SkX-MAX with all proton capture Q-values set to the highest values within the error bars of our
binding energy predictions. A similar set of calculations has been performed for the mass extrapolationsof Audi and
Wapstra 1995 [7] (AW95) and are labeled AW, AW-MIN, and AW-MAX. Fig. 5 shows the X-ray burst light curve,
the nuclear energygenerationrate, the abundances ofthe most importantwaiting point nuclei andthe hydrogenand
helium abundances as a function of time for all our calculations. As an example, Fig. 6 shows the time integrated
reactionflow correspondingto the SkX calculation. While the αp andrpprocesses below 56Ni are responsible for the
5
rapid luminosity rise at the beginning of the burst, processing through the slow waiting points 64Ge, 68Se, 72Kr and
the operation of the SnSbTe cycle (indicated by the 104Sn abundance) lead to an extended burst tail. The rpprocess
from 56Ni to 64Ge, and the slowdown at 64Ge lead to a pronounced peak in the energy generation rate around 50 s
after burst maximum. In principle the other waiting points have a similar effect, but the corresponding peaks in the
energy production are much wider and therefore not noticeable.
Fig. 7 compares X-ray burst light curves for different assumptions on nuclear masses. Generally, lower proton
capture Q-values enhance photodisintegrationand favor the waiting point nuclei in local equilibria. Both effects lead
to a slower reaction flow and therefore to less luminous but longer lasting burst tails. Even though the uncertainties
in our new mass predictions are significantly smaller than in AW95, they still allow for a burst length variation from
150 – 250 s and a luminosity variation of about a factor of 2 (SkX-MIN and SkX-MAX). The lower limit Q-value
calculationwith AW95masses(AW-MIN) is similarto our lowerlimit (SkX-MIN), but the largeruncertainties inthe
AW95 masses lead to large differences in the upper limits (SkX-MAX and AW-MAX) and would imply significantly
shorter bursts with much more luminous tails (AW-MAX). However, some of the large proton capture Q-values in
AW-MAXandto alesserdegreeinSkX-MAX arealreadyconstrainedby the experimentson69Brand73Rb. If those
constraintsaretakenintoaccountoneobtainstheAW-MAXEXPandSkX-MAXEXPcalculationsrespectively,which
are also shown in Fig. 7. The SkX-MAXEXP and AW-MAXEXP light curves are very similar.
The dependence of the light curves on the choice of proton capture Q-values can be understood entirely from the
changes in β-decay and proton-capture branchings of the main waiting points 64Ge, 68Se, and 72Kr shown in Table
I. The calculations with the lower limits on protoncapture Q-values (SkX-MIN and AW-MIN) do not differ much as
theyallpredictthatprotoncapturesdonotplayarole. However,fortheupperlimitssizableprotoncapturebranches
occur and lead to significant reductions in the lifetimes of the waiting points. In our upper limit (SkX-MAX) we
obtain26%protoncaptureon68Se (via2pcapture)and86%protoncaptureon64Ge, while protoncaptureson72Kr,
with 8%, play only a minor role. These branchings become even larger for the AW95 upper limit calculations (AW-
MAXEXP and AW-MAX). Note that β decay of 60Zn is negligible (See Table I) because proton capture dominates
for the whole range of nuclear masses considered here.
Theimportanceoftheone-protoncaptureQ-valuesinthedeterminationoflifetimesforrp-processwaitingpointshas
beendiscussedextensivelybefore[3]. Thisimportanceisclearlyexpressedbythelargechangesinbranchingratiosand
light curves when experimental constraints (which only exist for one-proton separation energies) are imposed on the
AW-MAXcalculationsleadingtoAW-MAXEXP(Fig.7andTable I). However,thetwo-protoncaptureQ-valuescan
be equally important. For example,the protoncapturebranchingon68Se changesby anorderofmagnitude from2%
inAWto 15%inAW-MAXEXP.This changeis entirelydue to the changeinthe 70Krprotonseparationenergyfrom
1.86 MeV in AW to 2.4 MeV in AW-MAXEXP as the proton capture Q-value on 68Se is very similar (only 0.05 MeV
difference). The reason for this sensitivity is the onset of photodisintegration of 70Kr that depends very sensitively
on its protonseparationenergy. As soon as temperatures aresufficiently high for 70Kr(γ,2p)68Se to play a role, 68Se,
69Br, and 70Kr are driven into a local (p,γ)-(γ,p) equilibrium. With rising temperature the proton capture on 68Se
drops then quickly to zero,because the temperature independent and slowβ decay of 70Kr in Eq. 5 cannot provide a
substantialleakageoutoftheequilibrium. ThisisdifferentfromthesituationatlowertemperaturesdescribedbyEq.4
wherealowerequilibriumabundanceof69Brathighertemperaturescanbe somewhatcompensatedbytheincreasing
protoncapturerateon69Br. ThiseffectisillustratedinFig.8whichshowsthelifetimeof68Seagainstprotoncapture
and β decay as a function of temperature for different choices of proton capture Q-values. The lifetime equals the
β-decay lifetime for low temperatures because of slow proton capture reactions, and at high temperatures because of
the photodisintegrationeffect discussedabove. Forthe AWmasses,the lowprotonseparationenergyof70Krleadsto
strongphotodisintegrationalreadyattemperaturesaround1.15GKbeforeprotoncapturescanplayarole. Therefore
proton captures never reduce the lifetime significantly. For AW-MAXEXP, the only change is a larger 70Kr proton
separationenergyof2.4MeV.Though69Brisunboundby500keV,protoncapturescanreducethelifetimeof68Seby
about a factor of two around1.4 GK before photodisintegrationsets in and starts inhibiting further protoncaptures.
Thiscanbecomparedwiththe upperlimits ofourpredictionsforprotonseparationenergies(SkX-MAX). Thelarger
proton separation energy of 69Br allows an onset of proton captures at slightly lower temperatures, but the lower
proton separation energy of 70Kr leads also to an onset of photodisintegration at somewhat lower temperatures thus
effectivelyshiftingthedropinlifetimebyabout0.1GK.Notethatitisnotonlytheamountoflifetimereduction,but
also how well necessary conditions match the actual conditions during the cooling of the X-ray burst that determine
the role of proton captures and therefore the overall time scale of the rp process. As Fig. 8 shows, both depends
sensitively on the nuclear masses.
Along-standingquestionishowthenuclearphysics,andinparticularthepropertiesofthelong-livedwaitingpoints
64Ge,68Se,and72Kraffecttheend-pointoftherpprocess. EvenforourlowestprotoncaptureQ-valueswhereproton
captures on 68Se and 72Kr become negligible, we still find that the rp process reaches the SnSbTe cycle [6]. Fig. 9
shows the final abundance distribution for the two extreme cases - our calculation with the slowest (SkX-MIN) and
the fastest (AW-MAX) reaction flow. In both cases, the most abundant mass number is A = 104, which is due to
6
accumulation of material in the SnSbTe cycle at 104Sn. The main difference between the abundance patterns are the
abundances that directly relate to the waiting points at A=64, 68, and 72 and scale roughly with the waiting point
lifetime. In addition, for AW-MAX nuclei in the A = 98–103 mass range are about a factor of 3 more abundant
because of the faster processing and the depletion of A=64, 68, and 72.
V. SUMMARY AND CONCLUSIONS
Wehavemadeanewsetofpredictionsforthemassesofproton-richnucleionthebasisofthedisplacementenergies
obtained from spherical Hartree-Fock calculations with the SkX Skyrme interaction [12,13]. SkX provides a
csb csb
large improvement in the displacement energies over those obtained with other Skyrme interactions via the addition
of a one-parameter charge-symmetry breaking component [13]. A comparison with the experimental displacement
energies measured in the mass region A=41-59 indicates that the accuracy of the calculated displacement energies is
about 100 keV. We thus use this as a measure of the uncertainty expected for the higher mass region of interest in
this paper. Experimental masses for some proton-rich nuclei in the mass region A=60-70 will be required to test our
predictions. At the upper end, we may expect some deviation due to the very deformed shapes which involve the
excitation of many pf-shell nucleons into the g (sdg) shell which go beyond our spherical approach. In addition
9/2
to the application to the rp-process, we have discussed the implication of the present model for the proton drip line.
The most promising candidates for diproton emission are 64Zn, 59Ge, 63Se, 67Kr and 71Sr.
Our rp-process calculations based upon the masses obtained in the present model and those obtained from the
Audi-Waptra mass extrapolations demonstrate clearly the sensitivity of X-ray burst tails on nuclear masses at and
beyond the N = Z line between Ni and Sr. Such a sensitivity on the Q-values for proton capture on 64Ge and 68Se
has been pointed out before by Koike et al. [5] based on a similar X-ray burst model. However, Koike et al. [5] used
a limited reaction network including only nuclei up to Kr. As we show in this paper, this is not sufficient for any
assumption on nuclear masses, and as a consequence we find very different light curves and final abundances.
Our new calculation leads to tighter constraints on proton capture Q-values as compared with the AW95 mass
extrapolations (see Fig. 7). The first radioactive beam experiments including the nonobservation of 69Br and 73Rb
have also begun to provide important constraints. If those experiments are taken into account, our new predictions
do not lead to substantially tighter limits, with the exception of the proton capture on 64Ge, where no experimental
upper limit on the proton capture Q-value exists. Our new calculations increase the minimum β branching at 64Ge
by an order of magnitude from 1% to 14%, leading to a lower limit of the average 64Ge half-life in the rp process of
12.6 s instead of 0.9 s. As a consequence,we predict a smoothand continuous drop in the light curve during the first
30–40 s after the maximum, as opposed to the hump predicted with AW-MAX.
However, uncertainties in the mass predictions are still too large to sufficiently constrain the light curves and to
determinetherolethatprotoncapturesplayinthereductionofwaitingpointlifetimes. Whilewefindthatwithinthe
errorsofourmasspredictionsprotoncaptureon72Krisnegligible,ourpredictedaverageprotoncapturebranchingsfor
64Ge and68Se still covera largerangeof 0.5%–86%and 0.0%- 26%respectively (ofcourse this is a model-dependent
result - for example more hydrogen or a higher density could strongly increase the proton capture branches). To
a large extent this is because of the large uncertainties in the masses of N = Z nuclei 64Ge (measured: 270 keV),
68Se (AW95 extrapolated: 310 keV), and 72Kr (measured: 290 keV) [7] that cannot be determined with the method
presented here. In addition, uncertainties in the masses of mirror nuclei increase the errors for 73Rb (170 keV) and
70Kr (160 keV) substantially beyond the ≈100 keV accuracy of our predicted Coulomb shifts. Overall,this results in
typical uncertainties of the order of 300 keV for several of the critical proton capture Q-values.
To summarize, uncertainties in the masses of the nuclei that determine the proton capture branches on 64Ge
and 68Se represent a major nuclear physics uncertainty in X-ray burst light curve calculations. The relevant nuclei
are listed in the upper part of Table II together with the currently available mass data and their uncertainties.
The proton capture branches on 60Zn and 72Kr are of similar importance, but are sufficiently well constrained by
current experimental limits and theoretical calculations. However, both the experimental and the theoretical limits
are strongly model dependent. Therefore, improved experimental mass data would still be important to confirm the
present estimates. These nuclei are listed in the lower part of Table II. As discussed in Sec. III there is experimental
evidence indicating proton stability of all the nuclei listed, except for 69Br and 73Rb, which are probably proton
unbound. Mass measurements of the proton bound nuclei could be performed with a variety of techniques including
ion trap measurements, time of flight measurements, or β decay studies. Recent developments in the production of
radioactive beams allow many of the necessary experiments to be performed at existing radioactive beam facilities
such as ANL, GANIL, GSI, ISOLDE, ISAC, and the NSCL. Mass measurements of the proton unbound nuclei 69Br
and 73Rb require their population via transfer reactions from more stable nuclei, or by β decay from more unstable
nuclei. Both are significantly more challanging as much higher beam intensities or the production of more exotic
7
nuclei are required, respectively.
Of course, burst timescales depend sensitively on the amount of hydrogen that is available at burst ignition. The
more hydrogen that is available the longer the rp process and the longer the burst tail timescale. In this work we
use a model with a large initial hydrogen abundance (close to solar) to explore the impact of mass uncertainties on
X-ray burst light curves. This allows us to draw conclusions on the uncertainties in predictions of the longest burst
timescales and the heaviest elements that can be produced in X-ray bursts. The former is important for example in
light of recent observations of very long thermonuclear X-ray bursts from GX 17+2 [30], the latter for the question
of the origin of p nuclei discussed below. Nevertheless we expect a similar light curve sensitivity to masses for other
models as long as there is enough hydrogen for the rp process to reach the A = 74−76 mass region. In our one
zone model we find that this requires about a 0.35-0.45 hydrogen mass fraction at ignition. Even though the burst
temperatures and densities vary somewhat with the initial conditions we find shorter, but otherwise very similar
reaction paths governed by the same waiting point nuclei. For bursts with initial hydrogen abundances below ≈ 0.3
the rpprocessdoes notreachthe A=60−72massregionanymoreandthe massuncertaintiesdiscussedinthis work
become irrelevant.
Observedtype I X-ray bursts show a wide variety of timescales ranging from 10 s to hours. Our goal is to improve
the underlying nuclearphysics so thatthe observedburst timescalescanbe used to infer tightconstraintsonignition
conditions in type I X-ray bursts such as the amount of hydrogenavailable for a givenburst. Such constraints would
be extremely useful as they could, for example, lead to constraints on the impact of rotation and magnetic fields on
the fuel distribution on the neutron star surface as well as on the heat flux from the neutron star surface [29,31].
Our results indicate that without further theoretical or experimental improvements on nuclear masses it will not be
possible to obtain such tight, quantitative constraints.
Nevertheless,somequalitativeconclusionscanalreadybedrawnonthebasisofournewmasspredictions. Ournew
resultsprovidestrongsupportforpreviouspredictionsthatthe rpprocessinthe A=64−72massregionslowsdown
considerablyleadingtoextendedbursttails[6]. Asaconsequence,thelongburstsobservedforexampleinGS1826-24
[31] can be explained by the presence of large amounts of hydrogen at ignition and can therefore be interpreted as a
signature of the rp process.
Even for our lowestproton capture Q-values,when 68Se and 72Kr slow down the rp process with their full β-decay
lifetime the rp process still reaches the SnSbTe cycle. Clearly, such a slowdown of the rp process does not lead to a
premature termination of the rp process as has been suggested previously (for example [2]), but rather extends the
burst time scale accordingly. As a consequence we find that hydrogen is completely consumed in our model.
However, a slower rp process will produce more nuclei in the A = 64–72 range and less nuclei in the A = 98–103
mass range. Interestingly, among the most sensitive abundances beyond A = 72 is 98Ru, which is of special interest
as it is one of the light p nuclei whose origin in the universe is still uncertain. p nuclei are proton rich, stable nuclei
that cannot be synthesized by neutron capture processes. While standard p process models can account for most
of the p nuclei observed, they cannot produce sufficient amounts of some light p nuclei such as 92,94Mo and 96,98Ru
(for example [32]). Costa et al. [33] pointed out recently that a increase in the 22Ne(α,n) reaction rate by a factor
of 10-50 above the presently recommended rate could help solve this problem, but recent experimental data seem to
rule out this possibility [34]. Alternatively, X-ray bursts have been proposed as nucleosynthesis site for these nuclei
[3,6]. An accurate determination of the 98Ru production in X-ray bursts requires therefore accurate masses in the
A = 64−72 mass range. Further conclusions concerning X-ray bursts as a possible p process scenario have to wait
for future self-consistent multi-zone calculations with the full reaction network,that include the transfer of the ashes
into the interstellar medium during energetic bursts.
Support for this work was provided from US National Science Foundation grants number PHY-0070911and PHY-
95-28844.
TABLE I. Branchings for proton captures on the most important waiting point nuclei for different mass predictions from
AW95 (AW) and this work SkX.These branchingsare thetime integrated averages obtained from ourX-ray burst model.
Waiting point SkX SkX-MIN–SkX-MAX AW-MIN–AW-MAX AW-MIN–AW-MAXEXP
60Zn 95% 91% - 97% 83% - 98% 83% - 99%
64Ge 30% 0.5% - 86% 0.0% - 98% 0.0% - 99%
68Se 0.5% 0.0% - 26% 0.0% - 74% 0.0% - 15%
72Kr 0.0% 0.0% - 8% 0.0% - 87% 0.0% - 8%
8
TABLE II. Nuclei for which more a accurate mass would improve the accuracy of rp process calculations in type I X-ray
bursts. The upper part of the table lists nuclei for which the current uncertainties lead to large uncertainties in calculated
burst time scales. The lower part of the table lists nuclei, for which accurate masses are important, but current estimates of
the uncertainties do not lead to large uncertainties in rp process calculations. Nevertheless, an experimental confirmation for
the masses being in the estimated range would be important. Within each part, the nuclei are sorted by uncertainty, so a
measurement of thetop ranked nuclei would be most important. For each nucleuswe list either the experimental mass excess
(Exp) ( [7] and [18] for 70Se) or thetheoretical mass excess (SkX)calculated in this work in MeV.
Nuclide Exp SkX
68Se -54.15 ± 0.30a
64Ge -54.43 ± 0.250
70Kr -40.98 ± 0.16
70Seb -61.60 ± 0.12
65As -46.70 ± 0.14
69Br -46.13 ± 0.11
66Se -41.85 ± 0.10
72Kr -54.11 ± 0.271
73Rb -46.27 ± 0.17
73Krb -56.89 ± 0.14
74Sr -40.67 ± 0.12
61Ga -47.14 ± 0.10
62Ge -42.38 ± 0.10
a Theoretical estimate from AW95.
b Mirror to an rp process nucleus - a more accurate mass measurement could reduce the error in the mass prediction
for the proton rich mirror nucleus by more than 30%.
9
[1] R.K. Wallace, S. E. Woosley, Ap.J. Suppl.45, 389 (1981).
[2] L. VanWormer et al., Ap.J. 432, 326 (1994).
[3] H.Schatzet al., Phys. Rep.294, 167 (1998).
[4] H.Schatz, L. Bildsten, A. Cumming, and M. Wiescher, Ap. J. 524, 1014 (1999).
[5] O.Koike, M. Hashimoto, K. Arai, and S.Wanajo, Astron. Astrophys.342, 464 (1999).
[6] H.Schatzet al., Phys. Rev.Lett. 86, 3471 (2001).
[7] G. Audi,A. H.Wapstra, Nuclear Physics A595, 409 (1995).
[8] B. A. Brown, Phys. Rev.C42, R1513 (1991).
[9] W. E. Ormand,Phys. Rev.C53, 214 (1996).
[10] W. E. Ormand,Phys. Rev.C55, 2407 (1997).
[11] B. J. Cole, Phys.Rev.C54, 1240 (1996).
[12] B. A. Brown, Phys. Rev. C58, 220 (1998).
[13] B. A. Brown, W. A. Richterand R.Lindsay, Phys.Lett. B 483, 49 (2000)
[14] A.Volya, B. A. Brown and V. Zelevinsky,Phys.Lett. B 509, 37 (2001).
[15] W. A.Richter,M. G. van derMerwe, R. E. Julies and B. A. Brown, Nucl.Phys. A523, 325 (1991).
[16] M. Hjorth-Jensen, T. T. S.Kuo and E. Osnes, Phys. Reports261, 125 (1995).
[17] S.Raman, C. W.Nestor and P.Tikkanen, Atomic Data and Nuclear Data Sheets,to be published (2001).
[18] B. E. Tomlin et al., Phys. Rev.C 63, 034314 (2001)
[19] B. Blank et al., Phys.Rev.Lett. 74, 4611 (1995).
[20] J. Winger et al., Phys.Lett. B 299, 214 (1993).
[21] J. D. Robertson et al., Phys. Rev.C 42, 1922 (1990)
[22] R.Pfaff et al., Phys. Rev.C 53, 1753 (1996).
[23] M. F. Mohar et al., Phys. Rev.Lett. 66, 1571 (1991).
[24] A.Jokinen et al., Z.Phys. A 355, 227 (1996).
[25] Z. Janas et al., Phys.Rev. Lett.82, 295 (1999).
[26] L.Bildsten,inTheManyFaces ofNeutron Stars,editedbyA.Alpar,L.Bucceri,andJ.VanParadijs(Dordrecht,Kluwer,
1998), pp.astro–ph/9709094.
[27] P.M¨oller, W.D. Myers, and J.R. Nic, At.Data Nucl.Data Tables 59, 185 (1995).
[28] T. Rauscherand F.-K.Thielemann, At.Data Nucl. Data Tables 75, 1 (2000).
[29] L. Bildsten, in Rossi 2000: Astrophysics with the Rossi X-ray Timing Explorer, March 22-24, 2000 at NASA’s Goddard
Space Flight Center, Greenbelt, MD USA 2000, p. E65.
[30] E. Kuulkerset al., submitted to Astron. Astrophys.astro-ph/0105386.
[31] A.Kong et al., Mon. Not. Roy.Astr.Soc. 311, 405 (2000).
[32] M. Rayet et al., Astron. Astrophys.298, 517 (1995).
[33] V.Costa et al., Astron. Astrophys.358, L67 (2000).
[34] M. Jaeger et al., Phys.Rev.Lett. 87, 202501 (2001).
10